Almost primes and the Banks-Martin conjecture (Q2308849)

Let \(\Omega(n)\) denote the number of prime factors of \(n\), counted with multiplicity. For \(k\geq 1\), the positive integer \(n\) is called \(k\)-almost prime is \(\Omega(n)=k\). The set \(\mathbb{N}_k\) consisting of all \(k\)-almost primes is a primitive set, i.e., no member of the set divides any other. \textit{P. Erdös} [J. Lond. Math. Soc. 10, 126--128 (1935; Zbl 0012.05202)] proved that \[ f(A):=\sum_{n\in A}\frac{1}{n\log n} \] is bounded uniformly over all primitive sets \(A\). Moreover, he conjectured that \[ f(A)\leq f(\mathbb{N}_1)=\sum_{p}\frac{1}{p\log p}=1.63661632\cdots. \] \textit{W. Banks} and \textit{G. Martin} [Integers 13, Paper A69, 10 p. (2013; Zbl 1285.11021)] conjectured that \(f(\mathbb{N}_k)>f(\mathbb{N}_{k+1})\) for all \(k\geq 1\). In the paper under review, the author proves that the conjecture of Banks and Martin is not true, by observing that the sequence \(f(\mathbb{N}_k)\) decreases for \(k\leq 6\) but then increases thereafter. More precisely, the author shows that \(f(\mathbb{N}_6)<f(\mathbb{N}_k)\) for all positive integers \(k\neq 6\). The author also proves the approximations \[ f(\mathbb{N}_k)=1+O(k^{-\frac{1}{2}+\varepsilon}), \] and \[ f(\mathbb{N}_k^*)=\frac{6}{\pi^2}+O(k^{-\frac{1}{2}+\varepsilon}), \] for any \(\varepsilon>0\), where \(\mathbb{N}_k^*\) is the set of squarefree \(k\)-almost primes.